AI Public Opinion Shows Concern, Not a Verdict on Risk

AI public opinion in an ABC News/Washington Post/Ipsos poll is markedly wary: 57% of Americans say AI is bad for society, compared with 42% who say it is good. That is a measure of public judgment, not a test of whether the risks people identify are already occurring or how likely they are.

AI public opinion points to concerns about security, work, and oversight

The poll describes more than general discomfort. Seventy-three percent say AI poses a high risk of hacking secure computer systems, and 65% say its growth poses a high risk of eliminating many jobs. Majorities also see a high risk that AI will reduce people’s ability to make decisions without it. Separately, two-thirds say the federal government is not doing enough to regulate AI’s growth and development.

Those findings suggest that concern is spread across distinct questions: security, employment, human agency, and government oversight. That breadth matters, but the figures do not tell us which concern drives the negative overall judgment, or whether people who oppose AI in general also favor a particular form of regulation. The poll establishes a pattern of reported views, not a single unified explanation for them.

Data center opposition grows closer to home

The infrastructure results make the gap between broad sentiment and local stakes especially visible. Sixty-six percent oppose building new data centers in the United States; asked about a new facility in their own community, 72% oppose it. The local figure is higher, though the summary does not establish why. It would be a mistake to treat that difference as proof of any one motive, such as concern about a facility’s effects or opposition to AI itself.

Still, this is a practical signal for anyone planning AI infrastructure: acceptance at a national level cannot be assumed to carry over to a particular community. That is an interpretation of the contrast, not a finding about what respondents would require from a proposed site. The poll gives no details here about which local concerns were measured or what conditions might change people’s views.

What the poll can—and cannot—establish about AI risk

The evidence gap is straightforward: reporting that many Americans perceive a high risk does not measure the actual frequency of AI-related cyberattacks, job elimination, or reduced decision-making ability. Nor does opposition to data centers, on its own, establish whether a specific facility is harmful or what policy would address public concerns. Those are different questions from the ones these results answer.

That distinction does not make the findings trivial. Public confidence affects the environment in which regulation and infrastructure decisions are made; dismissing concern because a poll cannot verify technical risks would confuse measurement with relevance. But the numbers should not be presented as proof that AI has already produced the harms respondents fear, either.

The narrower, more defensible claim is that the poll records substantial public concern about AI and meaningful opposition to new data centers, particularly in respondents’ communities. Whether those views reflect the risks of particular systems or sites remains a question for evidence beyond this poll.